ENCODING UNIQUE GLOBAL MINIMA IN NESTED NEURAL NETWORKS

被引：2
作者：
BARAM, Y [1 ]
机构：
[1] NASA,AMES RES CTR,MOFFETT FIELD,CA 94035
来源：
IEEE TRANSACTIONS ON INFORMATION THEORY | 1991年 / 37卷 / 04期
关键词：
NEURAL NETWORKS; NESTING; SPARSITY; CODING; GLOBAL MINIMA;
D O I：
10.1109/18.87008
中图分类号：
TP [自动化技术、计算机技术];
学科分类号：
0812 ;
摘要：
Nested neural networks are constructed from outer products of patterns over {-1,0,1}N, whose nonzero bits define subnetworks and the subcodes stored in them. The set of permissible words, which are network-size binary patterns composed of subcode words that agree in their common bits, is characterized and their number is derived. It is shown that if the bitwise products of the subcode words are linearly independent, the permissible words are the unique global minima of the Hamiltonian associated with the network.
引用
收藏
页码：1158 / 1162
页数：5
相关论文
共 8 条
[1]  
BARAM Y, 1988, NASA101032 TECH MEM
[2]  
BARAM Y, 1989, NASA102239 TECH MEM
[3]   ON THE CAPACITY OF ASSOCIATIVE MEMORIES WITH LINEAR THRESHOLD FUNCTIONS [J].
DEMBO, A .
IEEE TRANSACTIONS ON INFORMATION THEORY, 1989, 35 (04) :709-720
[4]   NEURAL NETWORKS AND PHYSICAL SYSTEMS WITH EMERGENT COLLECTIVE COMPUTATIONAL ABILITIES [J].
HOPFIELD, JJ .
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA-BIOLOGICAL SCIENCES, 1982, 79 (08) :2554-2558
[5]   HYPERCUBIC STRUCTURES IN ORTHOGONAL HOPFIELD MODELS [J].
HORN, D ;
WEYERS, J .
PHYSICAL REVIEW A, 1987, 36 (10) :4968-4974
[6]   OPTIMIZATION BY SIMULATED ANNEALING [J].
KIRKPATRICK, S ;
GELATT, CD ;
VECCHI, MP .
SCIENCE, 1983, 220 (4598) :671-680
[7]  
KOLMOS J, 1988, NEURAL NETWORKS, V1, P239
[8]   THE CAPACITY OF THE HOPFIELD ASSOCIATIVE MEMORY [J].
MCELIECE, RJ ;
POSNER, EC ;
RODEMICH, ER ;
VENKATESH, SS .
IEEE TRANSACTIONS ON INFORMATION THEORY, 1987, 33 (04) :461-482
1